The first change removes the run_maintenance step in install.sh.
That step is time consuming, memory consuming, more so as time goes
by. Eventually it will hit limits (e.g., number of locks acquired)
and fail. It happened in some repository already, and this step has
been removed there as a consequence. It is also an unnecessary step,
so we remove it.

Secondly, the refresh_journaling is a function that just delegates
to the _native variant, the only one possible these days. So in
this PR we swap them around and get rid of the wrapper.

db=> \sf journal.refresh_journaling
CREATE OR REPLACE FUNCTION journal.refresh_journaling(p_source_schema_name character varying, p_source_table_name character varying, p_target_schema_name character varying, p_target_table_name character varying)
 RETURNS character varying
 LANGUAGE plpgsql
AS $function$
BEGIN
  RETURN journal.refresh_journaling_native(p_source_schema_name, p_source_table_name, p_target_schema_name, p_target_table_name);
END;
$function$